Spectate Vlan for the target storage

[neo565656]

Hello,

I am new for veeam. One issue I have confused if I need to set.

Recently, I have started to use storeonce as backup repository
.
Since I create the a specific network card and vlan for the main vbr server to manage the backup job and it was a gateway server.

I m confused if I need to set my dedicated proxy server can have access to the storeonce too, since veeam mention the proxy as a data mover.

I can backup now but the speed is not fast as expected ( around 300mb/s, and I have 4 10 port with lacp enable between the server and storeonce. My target is to increase the transfer rate, and I m confusing this was because the everything queue on the main server)

I also consider to have jumbo frame to improve the speed of the backup.

Could anyone give some idea what I should improve?

[HannesK]

Hello,
and welcome to the forums.

The gateway server writes to the StoreOnce. So technically the proxy does not need access to StoreOnce. The point is, with "automatic selection", the machine hosting the proxy is also a gateway server (see table about how automatic selection works). So the answer is: it depends, whether you have automatic or static selection

For the performance: how many disks / machines do you back up in parallel? Are proxy and repository tasks high enough? What does the bottleneck analysis say (which values)?

Jumbo frames can help, but that's probably the last thing to look at.

Just to make sure: you use StoreOnce with Catalyst (not as SMB / NFS share)?

Best regards,
Hannes

[neo565656]

Hello Hannesk,

I have used a static gateway when I map the Storeonce to my vbr server.
as I understand, the automatic will let the vbr server to better allocate the resources for backup. due to the network segmentation, I have chosen static.

About the performance.
I have 60 machines that need to be backup, and I have almost 7TB for the total amount. of course, I'm using StoreOnce with Catalyst.
About the proxy, I have two windows servers with 12 CPU and 8GB ram to allocate.
As I check the log, it will be stuck on "Required backup infrastructure resources have been assigned", and all the backup up mention "Job will operate in low bandwidth catalyst mode"
that's why I'm still investigating the root cause of that.
About the bottleneck analysis, it said target.

One of the best practices I'm not following. As I read the recommendation from veeam, it said better to copy to the local disk and copy to the Storeonce, but for my case, I'm direct backup to the Storeonce.(disk resources not enough)

Thank you
Vanness

[HannesK]

Hello,
yes, dumb disk storage is cheaper and faster than deduplication appliances (except for use-cases with active-full backups), and that's why we recommend dumb disk storage with REFS or XFS file system.

Anyway:
1) Proxy CPU counts are more than enough, if tasks are configured correctly. RAM is not enough for the amount of CPUs. At least 24GB RAM would be needed for 12 CPU cores (please see my links above)
2) for 7TB, one proxy should be more than enough
3) bandwidth mode is often relatively irrelevant
4) without values, the bottleneck analysis is useless. Please see explanations I linked above. if it's always 99% target, then it's probably really the StoreOnce (if the gateway server is fine)
5) what model and which amount of disks you you have? 300MByte/s might simply be the limit that box can do (assumption, because you say that you have 7TB data).

Best regards,
Hannes

[neo565656]

Hi Hannes,

1) I will adjust it in shor
4) Let me read more about that
5) I 'm using storeonce 3620 and 7200rpm disk inside the storeonce.

Thank you
Vanness

[HannesK]

5) sounds like 6 disks according to this (information missing in your post). 300MByte/s sounds even faster than expected for such a small system